TOPICS IN
SURFACE
CHEMISTRY


THE IBM RESEARCH SYMPOSIA SERIES
Computational Methods in Band Theory
Editors: P.M. Marcus, J.F. Janak, and A.R. Williams
Computational Solid State Physics
Editors: F. Herman, N. W. Dalton, and T. R. Koehler
Sparse Matrices and Their Applications
Editors: D. J. Rose and R. A. Willoughby
Complexity of Computer Computations
Editors: R. E. Miller and J. W. Thatcher
Associate Editor: J. D. Bohlinger
Computational Methods for Large Molecules
and Localized States in Solids
Editors: F. Herman, A. D. McLean, R. K. Nesbet
Ion Implantation in Semiconductors
and Other Materials
Editor: Billy L. Crowder
Stiff Differential Systems
Editor: Ralph A. Willoughby
Optimal Estimation in Approximation Theory
Editors: Charles A. Micchelli and Theodore J. Rivlin
Topics in Surface Chemistry
Editors: Eric Kay and Paul S. Bagus

www.pdfgrip.com

TOPICS IN
SURFACE
CHEMISTRY
Edited by

Eric Kay
and

Paul S. Bogus
International Business Machines Corporation
San Jose, California

PLENUM PRESS . NEW YORK AND LONDON

www.pdfgrip.com


Library of Congress Cataloging in Publication Data
Main entry under title:
Topics in surface chemistry.
(IBM research symposia series)
Proceedings of an international symposium held Sept. 7-9, 1977, in the Federal
Republic of Germany.
Includes index.
1. Surface chemistry-Congresses. I. Kay, Eric. II. Bagus, Paul. III. Series: International Business Machines Corporation. IBM research symposia series.
QD506.A1T66

541'.3453

78-4888


ISBN-13: 978-1-4613-4005-8
e-ISBN-13: 978-1-4613-4003-4
DOl: 10.1 007! 978-1-4613-4003-4

Proceedings of an International Symposium
on Topics in Surface Chemistry
held in the Federal Republic of Germany, September 7-9, 1977
and sponsored by IBM

© 1978 Plenum Press, New York

Softcover reprint of the hardcover 15t edition 1978

A Division of Plenum Publishing Corporation
227 West 17th Street, New York, N.Y. 10011
All rights reserved
No part of this book may be reproduced, stored in a retrieval system, or transmitted,
in any form or by any means, electronic, mechanical, photocopying, microfilming,
recording, or otherwise, without written permission from the Publisher

www.pdfgrip.com


Preface

The papers in this volume were presented at an international
symposium on Topics in Surface Chemistry which was held in Bad
Neuenahr, West Germany., September 7-9, 1977. The symposium was
sponsored by IBM Germany.

It has been recognized for many years that our understanding
of bulk phenomena and their subsequent exploitation depends
largely on our ability to define correlations between microscopic
structure and the physical and chemical phenomena of interest.
The role played by surface phenomena in the overall behavior
of a material has been a subject for speculation for a long time,
but only during the last decade or so have experimental and theoretical tools been developed which make it possible to investigate
surface structure and related surface phenomena uniquely.
Numerous surface spectroscopies have been developed in
recent years intended to describe the geometric, vibrational and
electronic structure of a surface. Our present understanding of
surface, thin film and interfacial phenomena in solid state physics
owes much to these developments. In chemistry much of the interest
in surface science has come from the obvious implications to such
important and diverse fields as catalysis and corrosion. It takes
little imagination to recognize that there are many other areas
where advances in surface science can be brought to bear.
It was the purpose of this IBM sponsored conference to bring
together key scientists, particularly from Europe, who, though
active in quite diverse fields, appear to be asking related
questions about the role of surface structure and phenomena as
encountered in their particular fields of interest.
The motivation for the conference was to explore common
ground, especially in chemical aspects of surface and interfacial
phenomena. A conscious effort was made to intersect but also go
beyond topics covered by other chemically oriented surface conferv

www.pdfgrip.com



vi

PREFACE

ences, most of which had been motivated historically by the wide
interest in catalysis.
Five distinct fields were represented at the conference. The
sessions on "Fundamental Aspects of Surface Chemical Bonding" and
"Optical Excitations at Surfaces" examined recent progress in
understanding surface electronic and vibrational structure, including structure of sorbed species. Photoelectron Spectroscopy,
High Resolution Energy Loss Spectroscopy, Surface Raman Spectroscopy, as well as Attenuated Total Reflection and Surface Photovoltage Spectroscopy, were discussed. Both inorganic and organic
systems were considered. The session on "Atomic and Molecular
Scattering from Surfaces" took cognizance of the fact that ultimately any real understanding of surface chemistry at a gassurface interface must include a detailed description of energy
partitioning and the dynamics of surface scattering processes.
The session on "Surface Studies in Electrochemical Systems" dealt
with interfacial electrochemical phenomena and explored contemporary
chemical and physical approaches designed to study and control
electron transfer at such solid-liquid interfaces. The session
on "Ordered Array of Organic Molecules at Surfaces and Interfaces"
explored some very exciting chemical and physical characteristics
of complex organic molecules cast into well-defined monolayer
assemblies and also encountered in micellar structures.
Eric Kay
IBM San Jose Research Laboratory
Symposium Chairman

Paul Schweitzer
IBM Germany, Sindelfingen
Symposium Manager


www.pdfgrip.com


Contents

SURFACE STUDIES IN ELECTROCHEMICAL SYSTEMS
Reactions on Semiconductor Electrodes
R. Memming

1

Photoelectron Emission into Electrolytes • • • • • • • •
J. K. Sass

29

Formation of Surface Compounds on Electrodes
A. Bewick and M. Fleischmann

45

ORDERED ARRAYS OF ORGANIC MOLECULES AT
SURFACES AND INTERFACES

75

Monolayer Assemblies
D. M'dbius
Fast Radiation-Induced Processes in Micellar
Assemblies

M. Gdltzel
Polymerization of Diacetylenes in Multilayers
B. Tiecke and G. Wegner

103
121

ATOMIC AND MOLECULAR SCATTERING FROM SURFACES

--

---

Atomic and Molecular Scattering from Surfaces
Elastic Scattering
• • • • •
H. Wilsch

135

Reactive Scattering
M. Cavallini

161

Low Energy Ion Scattering • • • • •
U. Gerlach-Meyer and E. Hulpke

195


vii

www.pdfgrip.com


viii

CONTENTS

ASPECTS OF SURFACE CHEMICAL BONDING
Photoelectron Spectroscopy and Surface Chemistry
A. M. Bradshaw and D. Menzel

225

Impact of Surface Physics on Catalysis
T. Edmonds and J. J. McCarroll

261

........

Spectroscopy of Surface Vibrations
S. Andersson

291

OPTICAL EXCITATIONS AT SURFACES
Surface Photovoltage Spectroscopy


309

H. Lilth

Optical Spectroscopy of Surface Excitations
in Molecular Crystals and Monomolecular

Layers . • • • • . . • . . • . . . • • .

M. R. Philpott

329

Raman Spectroscopy at Surfaces
P. J. Hendra and M. Fleischmann

373

Index

403

www.pdfgrip.com


REACTIONS ON SEMICONDUCTOR ELECTRODES

R. Memming
Philips GmbH Forschungslaboratorium Hamburg
D2000 Hamburg 54, Germany

ABSTRACT
The essential charge transfer processes in reactions at semiconductor electrodes can be described on
the basis of an energy band model. The basic mechanisms
are described and examples for typical processes are
given. The main emphasis is put on photoeffects and
various processes induced by light absorption in the
semiconductor or in the electrolyte are discussed in
detail.
INTRODUCTION
Problems of the interface semiconductor-electrolyte were approached originally from two different
fields, from electrochemistry and from solid state
physics. Semidonducting properties played a certain
role already in an early stage of electrochemistry,
for instance in studies of oxide films on electrodes.
Since the electrical properties of semiconductors were
not known at that time systematic investigations of
electrode processes were not possible. Basic studies
in this field started shortly after single crystals
of semiconductors were available with well defined
electrical and optical properties.
In solid state physics the main interest was in
studies of semiconductor surfaces since surface effects
influence the electrical properties of many semiconductor devices.

www.pdfgrip.com


R. MEMMING

2


From electrochemical investigations not only information about the properties of semiconductor electrodes was obtained but also about the basic mechanisms
of charge transfer processes between electrodes and
molecules in the electrolyte which are not accessable
by using metals as electrode material. The studies in
semiconductor electrochemistry during the last 15 years
were not restricted to elemental semiconductors such as
germanium and silicon. Especially investigation with
large-band-gap semiconductors such as GaP, ZnO, SiC
and Sn02 (oand gaps between 2.3 and 3.8 eV) have contributea to a better understanding of the energy parameters and kinetics of electrochemical reactions. This
report is mainly focused on these processes. The principles of pure charge transfer processes are discussed
and illustrated by various examples.
POTENTIAL AND CHARGE DISTRIBUTION
The potential distribution at the semiconductorelectrolyte interface differs from that of a metal
electrode in so far that a potential drop does not
only occur across the Helmholtz double layer (UH) but
also across the space charge (Usc) below the semiconductor surface (Fig. 1). The formation of such a space
charge layer is due to the fact that the density of
charge carriers in the semiconductor is much smaller
than in a metal. Accordingly, the energy bands at the

>.
Ol

L..

Q)

C


Q)

Ev----r'"""'

pas. space
charge

Ec----~-';

EF - - - -

Ev----.-..

neg. space
charge

Fig. 1: Potential distribution at interface.

www.pdfgrip.com


3

REACTIONS ON SEMICONDUCTOR ELECTRODES

,I

I

:Q


c

"0a.

I

I
I

I
I

II
II

UH

I
11---"-_-

,

Ispace
Icharge II
Ilayer 1~lmholtz
1(-1f.1) ~
I layer

11-4Al

semiconductor

electrolyte

Fig. 2: Energy bands below the surface

semiconductor surface are bent upwards or downwards
depending on whether the space charge is positive or
negative (Fig. 2). The Fermi level is assumed to be
constant. Measuring the electrode potential versus a
reference electrode one obtains
(1 )

in which U ,U
and U are the potentials of the electrode, acr~ss the spac~ charge layer and across the
Helmholtz layer, respectively, whereas the constant
contains all other potentials occuring at the reference
electrode.
Varying the electrode potential by applying an external voltage the question arises which potential drop~
Usc' UH or both, are changed. This problem can be solvea
by capacity measurements. The capacity of a space charge
is defined as
(2)

The space charge (Qsc) - potential (Usc) dependence
can be derived from the Poisson equation

www.pdfgrip.com

..

R. MEMMING

in which the total charge p(x) is given by
p(x)

=

e[Nn - NA - n(x) + p(x)J.

(4)

The charge is determined b¥ all mobile carriers (electrons, n(x), and holes p(x») and by ionized donors Nn
and acceptors NA. The electron and hole densities ns
and Ps at the interface are related to the carrier densities no and PO in the bulk of the material by the
Boltzmann equatlons:
ns

=

no exp ( -

ps

=

Po exp (+

eU

sc) ,

~

(5a)

eU
k~C) .

(5b)

If equilibrium between electrons and holes exists
throughout the whole semiconductor then

in which ni is the inversion concentration, mostly a
very low value for large band gap semiconductors decreasing exponentially with the band gap.
Integrating the Poisson equation and using equations (2), (4) and (5) one obtains an exact relation
between space charge ca~acity esc and Usc. This equation is rather complex L1J and wlll not be discussed
here. In the exhaustion region (i.e. ns < no for n-type
and Ps < Po for p-type) this equation simpllfies to
(Mott-Schottky
equation)

(6)

In the case of large band gap semiconductors (E.g> 2 eV)
the exhaustion region is relatively large so th~t Eq.

(6) can be applied for determining the potential distribution.

www.pdfgrip.com


5

REACTIONS ON SEMICONDUCTOR ELECTRODES

LL

'"E

6

u

.,

u

"'III

~
~

2
,- "

-1.0

... "0

-0.5

0

.0.5

"" ,
",
.1.0

"

.1.5

electrode potential - V

Fig. 3: Capacity curves for n- and p-GaP.

A typical result as obtained with n- and p-type
GaP-electrodes ~2J is given in Fig. 3. According to
this figure 1/C$C varies linearly with the electrode
potential UE' Slnce the slope is identical to the theoretical value in Eq. (6) one has to conclude that any
variation of the electrode potential occurs across
space charge layer, i.e.

whereas the potential across the Helmholtz layer remains
unchanged. This result has been obtained with all semiconductors provided that the surface or the bulk was

not degenerated (Fermi level in the conduction or valence band). For further details it must be referred
to the literature [1, 3J.
From the capacity curves (Fig. 3) some further
important results can be derived as follows:
According to Eq. (6)
U

sc

-7

.+

kT ~ 0 for
e
~c.
Csc

-7

0, i. e.

www.pdfgrip.com


6

R. MEMMING

a)

-GaP

-I-

electrolyte--

f

C.SSeV
p -

b)

GaP

Ey - - - -

Fig. 4: Band bending of GaP at rest potential.

the extrapolation of the capacity curves in Fig. 3
yields an electrode potential, at which the energy
bands must be flat up to the surface (Usc = 0). These
so-called flat band potentials Ufb differ obviously
very much for n- and p-type GaP. Reconstructing now
the band bending for these two dopings from Fig. 3 for
a certai~ electrode ~otential (~.g. ~t UE = UR ) then
one obtalns a band plcture as glven ln Flg. 4. According to this model the distance of the Fermi level to
the band edges at the surface are equal for n- and
p-type, i.e. the position of the energy bands at the

surface is fixed independent of the electrode potential or
the doping of the electrode. At equilibrium the electrochemical potential (Fermi level) must be equal
throughout the whole system. Accordingly, varying the
doping Usc also changes. In order to keep the Fermi
level constant a corresponding change of the potential
drop at the metal-semiconductor contact on the back

www.pdfgrip.com


REACTIONS ON SEMICONDUCTOR ELECTRODES

7

side of the semiconductor crystal must change in exactly the opposite direction. At the flat band potential at which Usc = 0, one also knows the position of
Fermi level (equal now to the chemical potential of the
semiconductor) with respect to a reference electrode.
Since the Fermi level of a semiconductor is given by
EF

n
0
Ec + kT ln Nc

for n-type
(8)

and
EF

no
E - kT ln N
v
v

for p-type

(Nc and Nv density of states in conduction and valence
band) one can also determine the position of the conduction and valence band at the surface (E~ and E~).
Investigations with a large number of semiconductors have supported this result. According to such an
analysis the energy position of energy bands of various
semiconductors are given in Fig. 5 using the normal
hydrogen electrode (NHE) as a reference [1J. It should
be mentioned that these energy values can also be related to the vacuum level as it is usually done in
solid state physics (left hand scale in Fig. 5). hCcording to calculations by Lohmann [4J the energy of
NHE versus vacuum is of the order of
ENHE

=

(9)

-4.5 eV.

The values given in Fig. 5 depend on the pH-value of
the aqueous electrolyte. This result is due to the
fact that the potential drop across the Helmholtzlayer (UH) depends on pH. In one case this dependence
has been studied quantitatively [5J.
The position of energy bands of various semiconductors can be considerably different which is of great
importance for selecting a proper semiconductor for

studying a certain electrochemical process. As far as
electron transfer reactions are concerned another important result can be derived from measurements of the
potential distribution: Since any externally applied
potential occurs completely across the space charge
layer only the surface concentrations of electrons
and holes are varied according to Eqs. (5a) and (5b).

www.pdfgrip.com


8

R. MEMMING

or

(vacuum)

E

-3.0

-1.5

-3.5

-1.0

GaP
(n,P)

SiC
(n,P)
~

GaAsP
(~,P)

CdS
-4.0

-0.5

-4.5

-0

-5.0

+0.5

CdSe
(n)

~n)

Ti02
(n)

WO

~

~

(n?-Sn02~
(n)
~

-5.5
-6.0

+1.0

1.7 2.5
eV eV
~

3.2
eV

+1.5
+2D

-70

+2.5

-7.5

+3.0

3.2
eV

3.2
eV

~

-6.5

3.0
eV

0-

-

-

Eu 2 +/ 3 +

-

H2/H+

-

IFe(CN)6 13-/4 -

-

Fe 2 +/Fe 3 +

-

Ru(bipy)2+/3+

-

Ce 4 +/ 3 +

0-

3.8
eV

~

-8.0

+3.5

0-

redox
systems

Fig. 5: Position of energy bands at the surface of

various semiconductors at pH 1 •

www.pdfgrip.com


9

REACTIONS ON SEMICONDUCTOR ELECTRODES

CURRENT-POTENTIAL DEPENDENCES
The simplest way of getting information about reactions at semiconductor electrodes is the measurement
of a current-potential curve, an example given in Fig. 6
for n- and p-type GaP [6J. In this case the anodic process corresponds to the anodic dissolution of the material whereas the cathodic currents are due to hydrogen evolution. The cathodic current increases with potential for n-type whereas it is limited to a very low
value in the case of p-type. This result can only be
interpreted by assuming an electron transfer from the
conduction band to protons in the electrolyte, because

p-GaP

ELECTRODE POTENTIAL - V

-10

-05

/
UNDER ILLUMINATION

"'---"
-------- --

I

//

/
N

E

-1~
«

~

a)

-3

----

n - GaP
UNDER
+1

-lD
ELECTRODE
POTENTIAL - V

N

-1

E
u

:t

~

-2

b)

Fig. 6: Current potential curves for GaP in 1 N KC1.

www.pdfgrip.com


R. MEMMING

10

only very few electrons are available in the conduction
band of p-type. This is supported by the fact that the
cathodic current at p-type is increased by light absorbed by the semiconductor. The same arguments are
valid for the anodic dissolution. In this case, however, the process occurs via the valence band because
electrons can only be injected into the valence band
if holes are present. The small currents found with
n-type in the anodic range and with type during

cathodic polarization are limited by diffusion of the
corresponding minority carriers. In some cases these
currents remain at low levels even for very large polarizations. For instance, even for small band gap
semiconductors such as n-GaAs no current increase has
been observed up to 25 V [7J. The diffusion current
values and the potentials at which they suddenly increase is very much determined by the quality of the
crystal and the doping. In some cases avalanche breakdown (GaAs) [7J, in others tunneling processes through
the space charge (2nO [8J, Sn0 2 [9]) have been observed
According to Fig. 6 charge transfer process may
either occur via the conduction or the valence band.
The question arises, which factors determine whether
a certain process occurs via the conduction or the
valence band. More insight can be obtained by studying rather simple processes such as the electron transfer between semiconductor and redox system.
ENERGY LEVELS AND FERMI LEVEL IN ELECTROLYTES
Electrolytes containing an oxidation-reductor system
are characterized by the redox potential. Values of
this potential are usually given in the conventional
scale using the normal hydrogen electrode as a reference
point. Taking, however, the vacuum level as a reference
as it is common for solids, then the redox potential
corresponds to an energy required for transferring an
electron from a redox system in the electrolyte into
the vacuum. Accordingly, the electron energy is defined in the same way as in solids and one can define
a Fermi level of the redox system, EF el. At equilibrium
at the semiconductor-redox system int~rface the Fermi
levels of the solid (Er) and of the redox system
(EF,el) must be equal 1, 3J:
EF

=

EF,el .

Values are also given in Fig. 5.

www.pdfgrip.com


11

REACTIONS ON SEMICONDUCTOR ELECTRODES

The question arises now whether one can define or
derive also energy states for redox systems in a similar way as in semiconductors or metals. Certainly occupied and empty states are represented by the reduced
and oxidized species of a redox-system, respectively.
However, the energy position of these states can be
considerably different due to the strong interaction
of the redox system with the solvent. Each ion is
surrounded by a solvation shell. The interaction depends on the size and charge of the ion. If an electron
is transferred from the reduced species (Red), e.g. a
Fe 2 +-ion, into the vacuum then this process is followed
by a rearrangement or reorientation of the solvent molecules in the solvation shell. A similar rearrangement
is required for the reverse process. A certain energy,
the so-called rearrangement energy A is involved. The
complete cycle is given by

-

+e

Red solv , ox
-A

-A

~

-

-e
Red'solv,red I

OXsolv,ox

t-

A

( 11 )

OXsolv,red

in which the indices represent the state of the solvation shell, I the ionization energy and A the electron
affinity. Accordingly

I - A

= 2A

(12 )

•

It is assumed in this model that the rearrangement is
slow compared to the electron transfer (Frank-Condon
principle). Consequently the energy levels of the reduced (occupied) E~ed and oxidized species (empty)
Eg x are not equal. They differ from the Fermi level
EF,el by A as schematically shown in Fig. 7a.
The energy levels are not pure discrete levels
but are distributed over a certain energy range due to the
fluctuation of the solvation shell. The corresponding
distribution functions of the density of states are
given by
Dred

=

(E-E F,el -A) 2
exp 4kTA

(E-EF,el+ A)
exp 4kTA

2

(13 )

www.pdfgrip.com

12

R. MEMMING

o~-,.- vacuum

E

A

A
-

~.--.- EF,el

A

density of states

b)

a)

Fig. 7: Energy levels (a) and distribution (b) in redox
systems.

and are represented in Fig. 7. This energy picture for
redox systems has been proposed at first by Gerischer
[10J. The half width 6E~f2 of the distribution curves
are also determined by X:

1/2

6E 1 / 2 = 0.53 A

eVe

Typical values for inorganic redox systems are of the
order of 1 eV as determined by electrochemical methods
[11, 12, 13J, i.e. the energy levels of such a system
are spread over a considerable energy range.
The relative position of energy levels at the
interface can now be easily obtained using the condition that the Fermi levels are equal on both sides of
the interface. Examples for two redox systems of different normal potentials are given in Figs. 8a and c.
Redox systems of a large positive normal potential exhibit rather low lying energy levels in this energy
scheme, as also can be derived from Fig. 5 in which
various redox potentials are given.

www.pdfgrip.com


REACTIONS ON SEMICONDUCTOR ELECTRODES

13

E
E

Ec
E =-==-~.
F

-

at equilibrium -

Ec

E --F

E v- - -

a)

semiconductor

redox syst.

Dox

b)

anodic
~ polarisation

E redox system II

c)

d)
Ev- - -

Fig. 8: Relative position of energy levels at both
sides of the interface
a) and b) at equilibrium,
c) at anodic polarization,
d) at cathodic polarization.

CHARGE TRANSFER PROCESSES
Electron transfer and a corresponding current can
occur according to the energy picture if levels of equal energy exist on both sides of the interface. One can distinguish between various cases: In Fig. 9a, for instance,
the occupied and empty states of the redox system overlap with the conduction band. Applying anodic or cathodic
potentials corresponding currents can be observed. It
should be emphasized again that any externally applied
potential leads only to a variation of the band bending
whereas the relative position of the energy bands and
levels on both sides of the interface remain unchanged.
This is demonstrated for two cases in Figs. 8b and d.
The Fermi levels are not equal anymore, they differ now
by the externally applied potential ~, i.e.
(14 )

www.pdfgrip.com


R. MEMMING

14

One can distinguish now between various cases: In
Fig. 8a the empty (D Qx ) and the occupied states (Dred)

overlap with the conduction band. The anodic and cathodic currents are then given by
00

.+

"""

lC

i

c """

J

NcDreddE

(potential independent), (15a)

n s DoxdE

(potential dependent).

Ec

r

E
c

(15b)

In the first case electrons are transferred from occupied levels Dred to empty states in the conduction
band, the latter TIeing nearly identical to the total
density of states Nc at the lower edge of the conduction band since only few are occupied by electrons even
in n-type material. In this case the corresponding current is independent of the applied potential. The situation is different for a cathodic process, i.e. for an
electron transfer from the conduction band into empty

p-GoP
N

+1

E

~+100

<{

::J..
I

N

E

+0.5

~

~

L.
L.

.,.,,----------

<{

::J..
I

C

/

L.

:; -0.5

/

0.5

+1.0

+15

+50

::l

U

-10

electrode potentiol- V(vs SeE)

u

-10

o

-0.5

+1.0

-50

-100

Fig. 9: Current potent~al de~endence for n-Sn02 and
p-GaP in 5·10- M Fe +/Fe 3 + in 0.1 Ns S04.

www.pdfgrip.com


15

REACTIONS ON SEMICONDUCTOR ELECTRODES

states Dox of a redox system. In this case the current
depends on the density of electrons n at the surface
which is potential dependent accordiRg Eq. (5a); i.e.
the cathodic current increases exponential1 with ~.
An example is given for the system Sn0 2 /(Fe +/3+) in
Fig. 9a [11 J .

2

Similar equations are valid for a charge transfer
via the valence band. In this case the energy levels of
the redox system should overlap with the valence band
(Figs. 8c and d). The current equations are given by:
Ev

.+

lV

R:jJ

(potential dependent),

( 16a)

00

v oxdE

N D

(potential independent) . (16b)

00

Here an anodic current, i.e. electron transfer from Dred
into the valence band, can only occur if holes are present at the surface. The hole density is potential dependent according to Eq. (5b). For cathodic polarization,
electrons are transferred from the valence band to the
empty states of the redox system (D ox ), the current is
independent of the applied potential, an example is
given in Fig. 9b [6J.
It should be emphasized again, that currents determined by the surface concentrations of electrons (ns) or
holes (ps)
only increase with potential if the corresponding bulk concentrations (no and Po) are sufficiently large. Otherwise the currents wlll be diffusion limited.
As far as the current equations (Eqs. (15) and (16»
are concerned it should be mentioned that also other
theories have been developed such as by Marcus [14J and
Levich [15J and Dogonadze [16J. In all these cases the
reorientation energy A plays a dominant role and all of
these authors obtained the same dependence for i = f(A).
This is due to the fact that a harmonic oscillator picture has been used for the fluctuation of the solvation
shell. The description of a possible charge transfer by
the energy scheme discussed above has been introduced by
Gerischer [10J and is now well accepted in semiconductor
electrochemistry and proved by many results. On the basis
of this model one can predict very well which energy band

www.pdfgrip.com


16

R. MEMMING

will be involved if the position of energy levels in the
semiconductor and the redox is known (see Fig. 5). In
the case of redox systems in which two electron steps
are involved (e.g. H202 [2]) a prediction is more difficult because such a system has to be described by two
normal potentials [1, 2].

p-type

potential

+ redox syst.

n - type
part. anodic. c.

/'-'-'-'-'-'potential

i

l ___ ~~;al ~ed"d;ao

+redox syst.

._._._.\._._ ..../

!

~ ___
I
I

l'

Ured

,",,,.,

I
I

I

I

:

anod. dissol. :

:

redox syst.

I

I

i

L __________ -.J

Fig. 10: Current potential dependence for n- and p-type
semiconductor electrodes in the presence of
oxidizing agents (schematic).

www.pdfgrip.com


REACTIONS ON SEMICONDUCTOR ELECTRODES

17

Finally it should be mentioned that the current-potential curves due to a charge exchange between semiconductor and redox system may be distorted by the anodic
dissolution process occuring with many semiconductor
electrodes. An example is schematically shown in Figs.
10a and b. Assuming a redox process occuring via the
valence band, the cathodic currents (electron transfer
from Ev to Dox) are expected to be equal for n- and ptype materia~s. In the case of n-type, however, the
cathodic current occurs at much more negative potentials
than with p-type. This difference can be interpreted by
the different band bending found for n- and p-type at
a given electrode potential (see insert of Fig. 10).
Electron transfer from the valence band to the redox
system means hole injection. According to the sign of

the field within the space charge of n-type these
holes are pushed back towards the surface and are consumed for the anodic dissolution. Consequently, the small
current found with n-type is actually composed of two
partial currents; i.e. corrosion occurs now in a much
larger potential range than for p-type which is important
in etching processes.
PHOTOEFFECTS
One of the most interesting phenomena in semiconductor electrochemistry are the photoeffects. One photoeffect has already been mentioned in the discussion of the
basic current-potential curves in Fig. 6. If minority
carrier are involved in an electrode process then currents can be enhanced by light absorbed by the semiconductur as e.g. the anodic dissolution current for n-type
GaP (Fig. 6). It is interesting to note, however, that
the anodic current at n-type occurs during illumination
already at much more negative potentials that at p-type
in the dark. This problem can immediately be solved
after having discussed the origin of the photoeffect.
Light excitation in a semiconductor leads to the
formation of electron-hole pairs as indicated in Fig. 11.
They are separated by the electric field across the
space charge region. In the case of an upwards band
bending the holes are pushed towards the surface and
the electrons towards the bulk of the crystal. Under
open circuit conditions this leads to a decrease of
the electric field and consequently of the band bending.
This change can be detected as a corresponding shift of
the electrode potential (open circuit) towards cathodic
potentials. At sufficiently large light intensities

www.pdfgrip.com